A Review of Methods to Characterize Motor Unit Firing Properties

and Underlying Determinants

J. L. Dideriksen

, J. A. Gallego

, F. Negro

, A. Holobar

and Dario Farina

Department of Neurorehabilitation Engineering, Bernstein Focus Neurotechnology Göttingen,

Bernstein Center for Computational Neuroscience, University Medical Center Göttingen,

Georg-August University, Göttingen, Germany

Bioengineering Group, Spanish National Research Council (CSIC), Arganda del Rey, Spain

Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor, Slovenia

1 OBJECTIVES

This paper reviews traditional and novel techniques

for the characterization of motor unit firing

properties and the determination of their underlying

determinants.

These methods are becoming increasinly

important because of advances on techniques to

accurately identifiy the spike trains of several motor

units non-invasively (Holobar and Zazula, 2007,

Holobar et al., 2009), which enables the assessment

of the neural drive to muscle in an unprecedented

accurate fashion. It is further motivated by the fact

that traditional analysis using the surface

electromyogram (EMG) is largely influenced by a

number of intrinsic factors that limit the accuracy

that may be attained (Farina et al., 2004). Being the

most relevant among them the effect of the spatial

filter effect due to volume conduction of motor unit

action potentials through soft tissues to the skin

(where they are recorded, Merletti et al., 2008), the

influence of cross-talk among neighboring muscles

(Farina et al., 2004), and the effect of cancellation of

motor unit actions potentials (Keenan et al., 2004).

Throughout this review we employ the terms

motor unit or motor neuron according to common

usage in literature.

2 SIGNAL PROCESSING

METHODS

2.1 Motor Unit Firing Statistics

Statistical properties of the interval between motor

unit discharges provide the first relevant piece of

information when investigating the neural drive to

muscle. The histogram representing the distribution

of the time period between consecutive motor unit

discharges, termed inter-spike interval (ISI)

histogram has been widely used in the field. For

example, it has been shown that, in patients

suffering from some neurological diseases (e.g.,

Parkinson’s disease, Christakos et al., 2009), these

histograms exhibit abnormal patterns. Second and

third order distributions, which reflect the interval

between two and three consecutive

discharges, have

also proved to be useful in some contexts.

2.2 Motor Unit Synchronization

The development of techniques to accurately

estimate motor unit synchronization has received

considerable attention (see the reviews in Nordstrom

et al., 1992, and Negro and Farina, 2012) because of

the observed relation between common-stem

synaptic inputs and the increased possibility of

motoneurons firing simultaneously (Kirkwood and

Sears, 1978).

Most existing techniques are based on the

calculation of the cross-correlogram between pairs

of motor neurons, and the calculation of metrics

based on its characteristics. Relevant examples of

this are the Common Input Synchronization index

(CIS) proposed in (Nordstrom et al., 1992), which is

defined as the count of discharges in excess of

chance, and calculated as the area of the peak in the

cross-correlogram divided by its duration. Other

relevant synchronization metrics based on the cross-

correlogram are De Luca’s common drive index

(CDI, De Luca et al., 1982), and the synchronous

impulse probability (SIP, Datta et al., 1990).

Interestingly, it has also been shown that the

cumulative sum of the cross-correlogram permits

identifying the significant peak in the cross-

correlogram (Ellaway, 1978), and it is useful to

assess whether such peak is statistically significant

(Davey et al., 1986).

L. Dideriksen J., Gallego J., Negro F., Holobar A. and Farina D..

A Review of Methods to Characterize Motor Unit Firing Properties and Underlying Determinants.

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

The influence of the frequency of the synaptic input

has been a traditional concern as to the usage of

metrics to assess motor unit synchronization using

the cross-correlograms (Nordstrom et al., 1992). A

recent study by Negro and Farina demonstrated that

such influence significantly distorts the results in

healthy subjects, based on simulation and

experimental data (Negro and Farina, 2012). The

authors of that paper showed that a method based on

the activities of several motor units provides a better,

unbiased indicator of the properties of common

synaptic input to motor neurons (Negro and Farina,

2012), as explained below.

2.3 Common Synaptic Inputs to Motor

Neurons

As previously mentioned, animal studies proved that

the presence of an input common to a population of

motor neurons increases the possibility of such units

firing synchronously (Kirkwood and Sears, 1978).

However, the estimation of this input is largely

influenced by the statistics of input current and the

discharge rate of the motor neurons (Negro and

Farina, 2012). Indeed, higher discharges rates imply

a better sampling of the input current, and thus allow

a better reconstruction.

Under the assumption that groups of motor unit

spike trains (referred to as composite spike trains)

increases the average sampling rate of the common

input to motor neurons (Negro and Farina, 2011,

2011b), Negro and Farina showed that the coherence

between such composite spike trains provides the

best estimate of the strength and frequency of

common synaptic inputs to motor neurons. Notice

that coherence is the normalized Fourier transform

of the cross-correlation function, and that it is

independent of filter functions.

Interestingly, traditional methods for the

estimation of motor unit synchronization (see

Section 2.2), and thus of common input properties,

consisted in applying certain filters to the cross-

correlogram (e.g., De Luca et al., 1982, Nordstrom

et al., 1992), which corresponds, in the frequency

domain, to considering a certain frequency band of

the coherence spectrum (Negro and Farina, 2012).

This clearly shows that such estimators are

intrinsically influenced by the input frequency.

2.4 Corticospinal Coupling

The projection of supraspinal, typically cortical,

oscillations to the muscle has been commonly

investigated by computing the coherence between

the supraspinal signal and the surface EMG. This

technique has allowed demonstrating the existence

of cortical involvement during different type of

muscle contractions in healthy subjects (Conway et

al., 1995, Raethjen et al., 2008), and also in the case

of tremor (Volkmann et al., 1996).

Remarkably, since coherence is a linear

technique, the existence of significant coherence

between the supraspinal signal and the EMG implies

that the transmission is partly linear, despite the non-

linearity of the motor neuron transfer function

(Gerstner and Kistler, 2002, Negro and Farina,

2011). However, due to the intrinsic limitations of

the surface EMG, it is not possible to gather further

insight on how such projection actually occurs.

A recent study showed that linearity of the

transmission can only be achieved if the supraspinal

inputs mediating voluntary contraction are a

common synaptic input to the motor neuron pool

(Negro and Farina, 2011). This was demonstrated

based on two facts. The first was that sampling a few

motor neurons already provided significant

coherence, which implies that a few motor neurons

are able of transmitting the cortical input. The

second observation was that the accuracy of such

estimation (magnitude of the coherence) did not

further increase after a few motor neurons were

sampled, which indicates to a saturation in sampling,

only possible in the case of common inputs.

The demonstration of the linearity of the

transmission also has physiological implications,

because due to the non-linear properties of

interneurons (Gerstner and Kistler, 2002), linear

transmission implies that direct pathways mediate

voluntary contractions. Thus, such finding further

supports the relevant role of the corticospinal tract in

voluntary movement control (Lemon, 2008).

3 CONCLUSIONS

In this paper we have reviewed traditional and novel

methods to assess motor unit properties and their

underlying determinants. These methods are

becoming increasingly important since they permit

to assess motor unit activities with an unprecedented

accuracy, thereby enabling advances in basic

neuroscience, muscle physiology and motor control,

thanks to a more accurate characterization of the

neural drive to muscle

ACKNOWLEDGEMENTS

This work has been partly funded by the EU

Commission through grant ICT-2011-287739

[NeuroTREMOR].

REFERENCES

Christakos C., Erimaki S., Anagnostou E., Anastasopoulos

D. (2009). Tremor-related motor unit firing in

Parkinson’s disease: implications for tremor genesis.

Journal of Physiology 587, 4811-27

Conway B. A.,Halliday D. M., Farmer S. F.,

ShahaniU,Maas P,Weir AI & Rosenberg JR (1995).

Synchronization between motor cortex and

spinalmotoneuronal pool during the performance of a

maintained motor task inman. JPhysiol 489, 917–924.

Datta A. K., Stephens J. A. (1990) Synchronization of

motor unit activity during voluntary contraction in

man. The Journal of physiology 422: 397–419.

Davey N. J., Ellaway P. H., Stein R. B. Statistical limits

for detecting change in the cumulative sum derivative

of the peristimulus time histogram. J. Neurosci

Methods 1986;17:153–66.

De Luca C. J., LeFever R. S., McCue MP, Xenakis AP

(1982) Control scheme governing concurrently active

human motor units during voluntary contractions. The

Journal of physiology 329: 129–142.

Ellaway P. H. (1978). Cumulative sum technique and its

application to the analysis of peristimulus time

histograms. Electroencephalography and clinical

neuro- physiology

Farina D., Merletti R., Enoka R. M. (2004). The extraction

of neural strategies from the surface EMG. Journal of

Applied Physiology, 96(4), 1486-95

Gerstner W., Kistler K. Spiking Neuron Models.

Cambridge, UK: Cambridge University Press, 2002.

Holobar, A., Zazula D. (2007). Multichannel blind source

separation using convolution kernel compensation.

IEEE Transactions on Signal Processing, 55():4487-96

Holobar A., Farina D., Gazzoni M., Merletti R., Zazula D.

Estimating motor unit discharge patterns from high-

density surface electromyogram (2009). Clinical

Neurophysiologu 120, 551-62

Keenan K. G., Farina D., Maluf K. S., Merletti R., Enoka

R. M. (2005). Influence of amplitude cancellation on

the simulated electromyogram. Journal of Applied

Physiology 98(1), 120-31

Kirkwood P. A., Sears T. A. (1978) The synaptic

connexions to intercostal motoneurones as revealed by

the average common excitation potential. The Journal

of physiology 275: 103–134.Negro and Farina 2012

Lemon R. N. (2008). Descending pathways inmotor

control. Annu Rev Neurosci 31, 195–218.

Negro F, Farina D (2011) Linear transmission of cortical

oscillations to the neural drive to muscles is mediated

by common projections to populations of motoneurons

in humans. The Journal of physiology 589: 629–637.

Negro F, Farina D (2011b) Decorrelation of cortical inputs

and motoneuron output. Journal of neurophysiology

106: 2688– 2697.

Negro F, Farina D (2012). Factors influencing the

estimates of correlation between motor unit activities

in humans. PLoS One 7(9); e44894.

Nordstrom MA, Fuglevand AJ, Enoka RM. Estimating the

strength of common input to human motoneurons from

the cross-correlogram. J Physiol 1992;453:547–74.

Raethjen J., Govindan R. B., Binder S., Zeuner K. E.,

Deuschl G., Stolze H.(2008). Cortical representation

of rhythmic foor movements

. Brain Research 1236,

79-84

Volkmann J., Joliot M.,Mogilner A., Ioannides A. A.,

Lado F., Fazzini E., Ribary U. & Llinas R (1996).

Central motor loop oscillations in parkinsonian resting

tremor revealed by magnetoencephalography.

Neurology 46, 1359–1370.